unispeak

A common speaker for everyone

Comment

Inspiration

There are 13 different languages present in India, and it is literally impossible for any guy on the planet to know all of these languages. Thus giving rise to the problem of communication between different cultures. As soon as people move from one state to another, people face this huge uncomfortable situation where all the people in his surroundings don't understands the language that he has been familiar with since the beginning.

This gave rise to the idea of unispeak.

What it does

'Unispeak' is a smart application that uses machine learning algorithm to understand the context of any person speaking among one of around 110 languages that are supported by this application. This application takes the speech of any user and speaks it out in any one of the desired languages which has been choosen from the dropdown.

It has also been integrated with Telegram as a chat bot that takes as an input a voice message and on subsequent prompts about the input and the output language, gives you the text representation of the translated file.

This API is being highly scalable people will be able to connect other platforms as well according to their will. Integration of bots like Slack and Trello in the future will be of great boon for the companies where they have people from all across the world and they will be able to send voice messages and other people all across the world will be able to understand and feel what he's trying to express.

How we built it

Android

For building the android application we have used React native to build it, which essentially means exporting the application to iOS will also be very smooth and simple.

Server Side

For the syncing and processing of the Audio files, we have used node.js and we are using Google Speech API to understand and parse the data properly. Along with that we are using a few external libraries for properly parsing the file and uploading it to Google API. The final application has been deployed on a Heroku instance.

Challenges we ran into

Maintaining the audio file format and converting them without loosing there quality, since loosing the quality will make it difficult for the AI model to detect the speech.

Accomplishments that we're proud of

What we learned

  1. Making a signed apk while building with react-native
  2. Making a chat bot in Telegram with complexity of asking sequential questions.

What's next for unispeak

  1. Extend the functionality where users will be able to store recently translated texts in IPFS.
  2. Make more extensible bots for various platforms through which more people will be having the accessibility to speak in their own language.

Built With

Share this project:

Updates